Thermal Performance of Various Multilayer Insulation Systems Below ...

6 downloads 51 Views 479KB Size Report
Fermi National Accelerator Laboratory. P.O. Box 500, Batavia, Illinois 60510. April 1992. Presented at the Fourth Annual IISSC Conference, New Orleans, LA, ...
Fermi National Accelerator

Laboratory

FERMILAB-Conf4WS6

Thermal Performance of Various Multilayer Insulation Systems Below 80K W. Boroski, T. Nicol and C. Schoo Fermi National Accelerator Laboratory P.O. Box 500, Batavia, Illinois 60510

April 1992 Presented

e

at the Fourth Annual IISSC Conference, New Orleans, LA, March

Operated by Univerdes

Research Association Inc. under Contract No. DE-AW2-76CHOXOO

4-6, 1992.

MUJ the United States Depaiiment

of Energy

. Dsclaimer

This report was prepared us an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof

THERMAL PERFORMANCE OF VARIOUS MULTILAYER INSULATION

SYSTEMS BELOW 80K

W.N. Boroski, T.H. Nicol, and C.J. Schoo Fermi National Accelerator Laboratory’ P.O. Box 500 Batavia, Illinois 60510 USA

INTRODUCTION The SSC collider dipole cryostat consists of a vacuum shell operating at room temperature, two thermal shields operating near 80K and 20K respectively, and the The cryostat design superconducting magnet assembly operating near 4K’. incorporates multilayer insulation (MLI) blankets to limit radiant heat transfer into the 80K and 20K thermal shields. Also, an ML1 blanket is used to impede heat transfer through residual gas conduction into the 4K superconducting magnet assembly. A measurement facility at Fermilab has been used to experimentally optimize the thermal insulation system for the dipole cryostat. Previous thermal measurements have been used to define the 8OK ML1 system configuration and verify system performance*. With the 80K MLI system defined, the current effort has focused on experimentally defining the optimum insulation scheme for the 20K thermal shield. The SSC design specification requires that radiant heat transfer be limited to 0.093 W/m* at an insulating vacuum of 10” torr. The radiant heat transfer budget to 20K can be met without ML1 at insulating vacuums of approximately 10m6torr. Research indicates that very low heat transfer rates can be achieved between two low emissivity surfaces in the temperature regime below BOK, provided the insulating vacuum is significantly less than 10V6torr! However, this temperature region is very sensitive to residual gas conduction, with small pressure increases causing significant increases in total heat transfer. Under ideal laboratory or controlled conditions, low insulating vacuum levels are reasonably achieved and maintained. In contrast, insulating vacuums in large accelerators vary

* Operated by Universities Department of Energy

Research Association, Inc. under contract with the United States

significantly over time. For the SSC cryostat, variations in insulating vacuum will cause oscillations in refrigeration heat loads. Thus, an important advantage of installing ML1 on the 20K shield of the SSC cryostat is that the reflective layers serve as gas conduction barriers as well as radiation shields. These barriers will buffer the effects of transient conditions on the accelerator refrigeration system. The performance of several ML1 configurations has been evaluated under nominal and degraded insulating vacuum. For each system configuration, the warm boundary temperature remained fixed near 80K while the cold boundary temperature was varied from 10K to 50K. Since the SSC dipole cryostat employs aluminum thermal shields, all experimental arrangements included aluminum tape on the warm and cold boundaries. Heat transfer measurements were made with no ML1 installed on the cold boundary, and then with 10 and 5-layer ML1 blankets on the cold surface. Additionally, the effect of aluminum coating thickness on ML1 performance was studied by measuring the performance of systems with different aluminum coating thicknesses. EXPERIMENTAL

SETUP

Thermal measurements were made in the Heat Leak Test Facility (HLTF) illustrated in Figure 1. The HLTF as configured for 20K ML1 measurements has been previously described in detail, and thus only a brief overview is presented here.4 Test samples are installed around the cold plate, a cylindrical copper drum with an outer surface area of 0.36 m2. The cold plate outer surface is covered with aluminum tape to simulate the aluminum shield of the cryostat. ML1 test samples are installed as blankets around the cylindrical portion of the cold plate and as discs on the bottom section facing the 80K thermal shield. Blanket and disc layers are interleaved and taped to minimize the impact of the joint configuration. The cold plate is cooled by solid conduction through a heat flow meter. Heat flow meter accuracy is 2 1 mW near 20K, which corresponds to a heat flux of +2.8 mW/m2 given the cold plate surface area. The heat flow meter is attached to the anchor plate, which establishes the cold boundary temperature. Anchor plate temperature variation is achieved by varying the flow rate of cooling gas through tubing soldered to the cylindrical inner surface. A temperature controller is used to balance electrical power against this flow to maintain a constant temperature. Using the temperature controller and vent gas from 500-liter liquid helium dewars, constant temperatures over the range 10K to SOK are achieved. Encompassing the test sample is a copper shield operating near 80K. The inner surface area of the 80K shield is approximately 0.68 m2, with a radial distance of approximately 5 cm between the 80K shield and the cold plate. The 80K thermal shield is cooled through a bolted connection to the bottom of an LN, reservoir. The inner surface of the 80K shield is covered with aluminum tape to simulate the inner surface of the cryostat aluminum 80K shield. System pressure in the HLTF is measured with two Bayard-Alpert ionization gauges. Historically, the lowest insulating vacuums achieved in the HLTF were on the order of 6 x 10e6 torr. This level of vacuum had been acceptable for past measurements on 80K ML1 systems and cryostat mechanical support structures. However, it became clear early in the 20K ML1 measurement program that insulating vacuum on the order of 10m7torr would be necessary to reduce residual gas contribution to negligible levels. Consequently, modifications were made to the vacuum system during the course of the program: the pumping system was reconfigured and the vacuum port size enlarged for increased throughput. This explains why some of the data presented is over broader pressure ranges than others.

MEME TcwER*TULE “Acw cm AOISTER PLATE YI TEST YLPLE bEAT YTER mo PLATE cAL!aRAT!cu ICATER COLDPLATE

sa WRYU WlELD xK* IaoTF-LATE B*ImD-ALPERT “AcuLl GACX Fwc

1. Heat Leak Test FaciIity

for MU

measurements below 8OK

METHOD OF OPERATION Test samplesare installed with the HLTF 180” from the orientation shown in Figure 1. During initial pumpdown,heater tapeson the vacuumshell are powered to heat the apparatusand acceleratesystemoutgassing. The systemis evacuatedto 1 micron, rolled into operating position, and then backfilled with dry nitrogen gas. The systemis then evacuateda secondtime, the heater tapes turned o& and the vessels fiIIed with cryogens. This evacuation processoccurs over 3 working days. The Cst data points are taken under *good” Insulating vacuum Subsequentdata points are taken with the cold plate held near 2OK and the vacuum pump throttled to create ‘degraded” vacuum conditions. Finally, the vacuum is opened to the pump and the 2OK data point repeated to ensure that no shift occurred during degradedoperation MEASUREMENTRESULTS Table 1outhnesthe systemsevaluatedduring this program. The 6rst measurement with aIuminum tape on both boundary surfacessimulated a cryostatcondition where no MLI wasused. The secondsystemmodeledthe MU systemconfigurationcurrently specified for the SC 50 mm cryostat? This configuration consistsof 10 reflective layers each separated by three spacer layers of spunbonded polyester mat. The reflective layersare O&25mm thick Mylar 6Im aIumbdA on both sidesto a nominal coating thicknessof 600 angstroms. The triple spacer layers servesto maintain low

layer density, which is desirable at low temperatures where solid conduction is a dominant mode of heat transfer. The remaining systemswere included to study the performance impact of blanket thickness,sewnseams,and aluminum coatingthickness. Table L k%mmwy of insulatioa schemes evahut Law HW tm

-d

DNAt.aF Fc)

spurn Dma@m

tm= an* w-l

acar 2OK w An mrm?

rrpnr tzlzzy trwhm.Kl

voar)

The data in Table 1 correspond to steady-stateconditions at the noted vacuum levels. Thesepoints were selectedfor the table basedon insulating vacuumlevelsthat were closest to the cryostat designspecification: limit radiant heat flux into the 20K shield to 0.093W/m* at lo4 torr. The heat flux valuesmay appear higher than those found elsewherein the literature for MLI systemsnear 4K.’ This is due in part to the insulating vacuumat which this data was taken; gasconduction is a larger contributor at lob torr than at pressuresof 10’ torr and below. Thermal performanceof the three configurations (TZ,T3,T4) that incorporated600 angstrom DAM material is compared in Figures 2 and 3. At low pressures, performance is driven by the sewnseamgeometry; the steppedseamgeometryoffered the lowest heat flux, and was the only Mu systembelow design budget. As the pressure increases,the impact of the seambecomeslm sign&ant while the effect of blanket thicknessbecomesa factor. These issuesare dkcumed in subsequentsections. 10 -

10 IJn w/llepfed

warn

-lDi~~~DAU/ti~ht+m

-

1.003-07

5 lyrs W/shi~bl

l.OOE-06

urn

l.OOL-05

Vacuum(lorr) Fwe2

HeatflurthroughMLI.systasnur20K

l.OOE+O4

1. WE-03

10.000

1.000

-

IOlyn DAN/slrai@l ram

-

5 Iyn DAN/dmighlmm

0.100

0.010 l.OOE-07

l.OOE-06

l.OOE-05

l.OOE-03

l.OOS-04

Vacuum(Lorr) Fpe

3. Apparent thermal conductbity of MLI systems aear 2OK

Effect of SewnSeam on llrennal Performance Sewnseamsare used on all SSCMLI blankets to hold the many layers together during handling and cryostat installatiolzU As radiant heat transfer falls off rapidly below 8OK, solid conduction between MLJ layers becomes a more significant contributor to overall heat transfer. Thus, to study whether the conduction contribution through a seamat 20K has a signScant impact on overall performance, lO-layer blankets with straight and stepped seams(T4 and T2) were evaluated and compared. The straight seam was sewn through all 10 layers with minimal stitch tension; the stepped seamwassewnthrough 5 layers,offset by 7.6 cm, and then sewn through the remaining 5 layers. Measurementdata shown in Figure 4 illustrates that the steppedseamblanket has a lower heat flux to 2OKat low pressure; the heat flux through the straight seam exceedsthe design budget. As pressure increases,the performanceof both blankets is driven by gas conductionwith the seamcontribution negligible by comparison. As a result of this comparison,the design of the SSCdipole cryostat 20K MLJ blanket will continue to incorporate stepped sewn seams.

-straight

0.01 J 1.001-07

mewn

I I llllll 1.006-06

Ill1 1.0011-05

Vacuum (torr)

ltttttl l.OOW04

I lllttu 1.003-03

Effect of Blanket thickness of Thermal Performance Since radiation falls off rapidly below 8OK,the numberof reflective layersrequired to limit radiant beat transfer becomesless. In fact, if small heat fluxes are attainable between bare low.emissivitysurfaces,one might argue that the number of reflective layers is rather insignificant at these temperatures from a radiant heat transfer standpoint. The number of layers then become significant primarily during transient or upset conditions. TO study this, a comparison was madebetweena S-layerand 1CL layer MLl blanket of like geometry and layer density (T3 and T4). The singular difference betweentest sampleswasoverall thicknessdue to number of layers. Each blanket was comprisedof DAM sheetswith a coating thicknessof 600 angstromsper side. Figure 5 showsthat the performance of both blankets is comparable at low pressures,while heatflux through the S-layerblanket is approximately35% higher than through the IO-layerblanket as the pressure increasesto 104 tot-r. This data is in agreement with the theory that reflective sheets serveasgasconduction barriers, and supports the continueduse of Ml-layerblankets for the cryostat20K MLl system 10.000

0.010 1.00s07

l.OOE-06

l.OOE-05

l.OOE-04

l.OOE-03

Vacuum(tot-r) Figure 5. Effed of blankti thickaa

on overall thermal performance

Effect of Alundn~~ Coating lldclmess on Tbennal Performance It has been noted that the addition of h5LI at temperaturesnear 4K actually caused an increasein the rate of heat transfer? One explanation is based on the increased trammisstvity of thin aluminum Blms at very low temperatures.’ Basedon this theoretical prediction, the 2OKIvlLJ systemfor the SSCcryostat is comprked of reflective sheetsof DAM with aIuminum coating thicknessesof 600angstromsper side. By comparison,the 80K hfLl systememploys DAM sheetswith alumina coatingr of 350angstrom per side. To quantify the effect of aluminum coating thicknesson h4U performance at 2OK, measurementswere made on two blanketsof like geometry(T4 and T5). Each blanket was comprised of 10 layers of DAM with three layers of spunbonded polyester mat separatingeach reflective pair. The blanket layers were fastened together with parallel straight sewn seams. The singular difference wasthat one blanket wascomprisedof 600 angstromDAM, the secondwas comprisedof 350 angstrom DAM. Figure 6 illustrates that an increasein aluminum coating thickness

from 350 angstromsto 600 angstromshad no significant effect on overall thermal performance. As a result of this measurement,350 angstromDAh4 can be usedin all MLI systemsin the SSCcryostat

10

_--

F h E 8 N

g G 3 =

angstromejside 1

-350 angstroms/side

0.1 0.01 l.OOE-07

l.OOE-06

l.OOE-05

l.OOE-04

l.OOE-03

Vacuum(torr) Figure6.~~ofaluminumfoaringthicLnessonMLIpcrformana

CONCLUSION A number of insulation schemeshave been evaluatedbetween 80K and 20K to optimize the thermal insulation systemfor the SSC dipole cryostat. Experimental results verify that the lO-layer MLI blanket currently specified for the !SC dipole cryostat20K shieldmeetsthe designbudget at an insulatingvaeuumof lab torr. From the data, severalother key points are noted: The region near 20K is very sensitive to residual gas conduction. Insulating vacuummustbe below 10’ torr before the contribution of gas conduction to insulation performance becomesnegligible. At vacuum levels near la6 torr, the number of reflective layersis not as important as how the blanket is eot.@ured. Blankets fabricated with similar layer densitiesand fasteningmethodshad the sameperformance even though one blanket had double the number of reflective layers. ‘Ibe number of reflective la ers becomesimportant when insulating vacuum increases above 1J torr. The additional layers limit heat transfer by impeding residual gas conduction. Sewnseamsimpact the overall thermal performance of MLI blankets operating near 2OK. The steppedsewnseamgeometryspecified for the SSC2OK MLI systemallows the many layersto be handled as a single componentwhile still meeting the thermal designrequirements.

l

The ‘w of reflective films with 600 angstromthick aluminum coatings offer no significant improvement in performance over similar systems comprisedof films with 350 angstromthick coatings. This will allow the use of reflective films aluminized with coating thicknesses of 350 angstroms in all SSC cryostat blankets without compromising system performance.

ACKNOWLEDGEMENTS

The authors wish to thank M. Ruschmanand R. Kunzelmanfor their contributions in the developmentof the measurementfacility and their assistancein the acquisition of this data Gratitude is also expressedto D. Franks,R. Pletzer,A Yucel, and others for countlessdiscussionsregarding the experimental results, and to M. McAshan, R. Coombes,and E.G. Pewitt for their support of this work REFERENCES

1. T.H. Nicol, Design development for the 50 mm SuperconductingSuper Collider dipole cryostat, in: “Supercollider3,” Plenum Press,New York (1991), p. 1029. 2. J.D. Gonczy, W.N. Boroski, and RC. Niemann, Thermal performance measurementsof a 100 percent polyester MLI systemfor the SuperconductingSuper Collider: Part II, Laboratory Results (3OOK-80K),in: “Advancesin Cryogenic Engineering,” Vol. 35, Plenum Press,New York (1989). p.497. 3. E.M.W. Leung, et al., Techniquesfor reducing radiation heat transfer between 77K and 4.2K, in: “Advancesin CryogenicEngineering,”Vol. 25, Plenum Press,New York (1980), p. 489. 4. W.N. Boroski, RJ. Kunzebnan,M.K. Ruschman,and CJ. Schoo,Design and calibration of a test facility for MU thermal performance measurements below 8OK,presented at the Cryogenic Engineering Conference, June 11-14, 1991,Huntsville, Alabama 5. W.N. Boros&T.H. Nicol, and CJ. Schoo,Design of the multilayer insulation systemfor the SuperconductingSuper Collider 50 mm dipole cryostat, in: “Supercollider 3,” Plenum Press,New York (1991),p. 849. 6. I.E. Spradley,T.C. Nast, and DJ. Frank, Experimental studies of MLI systems at very low boundary temperatures,in: “Advancesin Cryogenic Engineering, VoL 35, Plenum Press,New York (1989). p. 477.